Chapter 7: Main Memory Rev. by Kyungeun Park, 2015

Chapter 7: Memory Management Background Swapping Contiguous Memory Allocation Paging Structure of the Page Table Segmentation Example: The Intel 32 and 64-bit Architectures Example: ARM Architecture

Objectives To provide a detailed description of various ways of organizing memory hardware To discuss various memory-management techniques, including paging and segmentation To provide a detailed description of the Intel Pentium, which supports both pure segmentation and segmentation with paging

Background Program must be brought (from disk) into memory and placed within a process for it to be run Main memory and registers are only storage CPU can access directly Memory unit sees only a stream of memory addresses + read requests, or address + data and write requests. We are focusing on sequence of memory addresses generated by the running program. Motivation Register are accessible within one cycle of the CPU clock. Whereas, main memory is accessed via a transaction on the memory bus. Completing a memory access may take many cycles of the CPU clock. Processor needs to stall, causing delay  intolerable! Hardware-based Remedy: Cache, fast memory, between CPU and main memory on the CPU chip without any operating-system control Additional Remedy: Protection of memory required to ensure correct operation  Base and Limit Registers provided by the hardware

Base and Limit Registers A pair of base (300040) and limit (120900) registers define the logical address space CPU hardware compares every address generated in user mode with the registers. Access to operating-system memory or other users’ memory results in a trap to the operating system.

Hardware Address Protection with Base and Limit Registers

Address Binding Processes (actually programs) on disk, ready to be brought into memory to execute from the input queue The first address of the user process need not be 0000. Logical address and physical address mapping Further, addresses are represented in different ways at different stages of a program’s life Source code addresses usually symbolic Compiled code addresses bind to relocatable addresses i.e. “14 bytes from beginning of this module” Linkage editor or loader will bind relocatable addresses to absolute addresses i.e. 74014 Each binding maps one address space to another

Binding of Instructions and Data to Memory Address binding of instructions and data to memory addresses can happen at three different stages Compile time: If memory location known a priori, absolute code can be generated; must recompile code if starting location changes Load time: Must generate relocatable code if memory location is not known at compile time Execution time: Binding delayed until run time if the process can be moved during its execution from one memory segment to another Need hardware support for address mapping (e.g., base and limit registers)

Multistep Processing of a User Program

Logical vs. Physical Address Space The concept of a logical address space that is bound to a separate physical address space is central to proper memory management Logical address – generated by the CPU; also referred to as virtual address Physical address – address seen by the memory-management unit Logical and physical addresses are the same in compile-time and load-time address-binding schemes Logical (virtual) and physical addresses differ in execution-time address-binding scheme Logical address space is the set of all logical addresses generated by a program Physical address space is the set of all physical addresses corresponding to the logical addresses

Memory-Management Unit (MMU) Hardware device that maps virtual to physical address at run time Many methods possible, covered in the rest of this chapter To start, consider simple scheme where the value in the relocation register is added to every address generated by a user process at the time it is sent to memory Base register now called relocation register MS-DOS on Intel 80x86 used 4 relocation registers The user program deals with logical addresses; it never sees the real physical addresses Execution-time binding occurs when reference is made to location in memory Logical address bound to physical addresses

Dynamic relocation using a relocation register (based register) Routine is not loaded until it is called Better memory-space utilization; unused routine is never loaded All routines kept on disk in relocatable load format Useful when large amounts of code are needed to handle infrequently occurring cases No special support from the operating system is required Implemented through program design OS can help by providing libraries to implement dynamic loading

Dynamic Linking Static linking – system libraries and program code combined by the loader into the binary executable program image waste of disk space and main memory Dynamic linking – linking postponed until execution time Stub, a small piece of code, is included in the image for each library-routine reference Stub is used to locate the appropriate memory-resident library routine When the stub is executed, it checks if routine is in processes’ memory address If not in address space, the program loads the routine into memory address space. Stub replaces itself with the address of the routine, and executes the routine Dynamic linking is particularly useful for language libraries  only one copy of the library code Consider applicability to patching system libraries Version information is needed System also known as shared libraries

Swapping A process can be swapped temporarily out of memory to a backing store, and then brought back into memory for continued execution Total physical memory space of processes can exceed physical memory e.g.) a multiprogramming environment with a RR CPU-scheduling When a quantum expires, swap starts and CPU-scheduling will be done. The quantum: large enough to allow reasonable amounts of computing to be done between swaps. Backing store – fast disk large enough to accommodate copies of all memory images for all users; must provide direct access to these memory images Roll out, roll in – swapping variant used for priority-based scheduling algorithms; lower-priority process is swapped out so higher-priority process can be loaded and executed System maintains a ready queue of ready-to-run processes which have memory images on disk Does the swapped out process need to swap back into the same physical addresses? Depends on the address binding schemes : Plus pending I/O to/from process memory space Assembly or load-time binding : the process cannot be easily moved to a different location. Execution-time binding : the process can be swapped into a different memory space. Plus pending I/O to/from process memory space

Schematic View of Swapping Dispatcher checks to see whether the next process in the queue is in memory. If it is not, and if no free memory region The dispatcher swaps out a process and swaps in target process. Then, reloads registers and transfers control to the selected process

Context Switch Time including Swapping Context switch time can then be very high 100MB process swapping to hard disk with transfer rate of 50MB/se 100 MB/50 MB per second = 2 seconds Swap out time of 2000 ms (2 seconds) Plus swap in of same sized process : 2000 ms (2 seconds) Total context switch swapping component time of 4000 ms (4 seconds) Major part of swap time is transfer time; total transfer time is directly proportional to the amount of memory swapped Can reduce swap time by swapping out small size of memory  by knowing how much memory really being used User must keep the system informed of any changes in memory requirements. A process with dynamic memory requirements need to issue system calls to inform OS of changing memory needs via request_memory() and release_memory().

Context Switch Time including Swapping (cont.) Make sure to swap out an idle process Plus consider pending I/O Swap out process P1 and swap in process P2 : I/O operation of process P1 might attempt to use memory now belonging to process P2  Two main solutions to this problem 1. never swap a process with pending I/O 2. execute I/O operations only into operating-system buffers (kernel space), then transfer between operating-system buffers to process memory then the process is swapped in. Modified versions of swapping are found on many systems (i.e., UNIX, Linux, and Windows) Swapping is normally disabled Swapping will start if more than threshold amount of memory allocated Swapping is disabled again once memory demand reduced below threshold

Swapping on Mobile Systems Not typically supported Flash memory based Small amount of space Limited number of write cycles Poor throughput between flash memory and CPU on mobile platform Instead use other methods to free memory if low iOS asks apps to voluntarily relinquish allocated memory Read-only data thrown out and reloaded from flash if needed Failure to free can result in termination Android terminates apps if low free memory, but first writes application state to flash for fast restart Both OSes support paging

Contiguous Allocation Main memory must support both OS and user processes Limited resource, must allocate efficiently Main memory usually divided into two partitions: Resident operating system, usually held in low memory with interrupt vector User processes then held in high memory In contiguous memory allocation, each process contained in single contiguous section of memory Relocation registers with limit register used to protect user processes from each other, and from changing operating-system code and data Base relocation register contains value of smallest physical address Limit register contains range of logical addresses – each logical address must be less than the limit register MMU maps logical address dynamically Relocation-register scheme provides an effective way to allow OS’s size to change dynamically  Flexibility transient operating-system code (uncommon device driver)  removed from memory Using transient code changes the size of the OS during program execution

Hardware Support for Relocation and Limit Registers

Contiguous Allocation (Cont.) Multiple-partition allocation The simplest method is to divide memory into several fixed-sized partitions Each partition may contain exactly one process. Degree of multiprogramming limited by number of partitions In the variable-partition scheme, a hole – initially, one large block of available memory; Eventually, a set of holes of various size, scattered throughout memory When a process arrives, it is allocated memory from a hole large enough to accommodate it Process exiting frees its partition, adjacent free partitions combined Operating system maintains information about: a) allocated partitions b) free partitions (hole) OS OS OS OS process 5 process 5 process 5 process 5 process 9 process 9 process 8 process 10 process 2 process 2 process 2 process 2

Dynamic Storage-Allocation Problem How to satisfy a request of size n from a list of free holes? First-fit: Allocate the first hole that is big enough Best-fit: Allocate the smallest hole that is big enough; must search entire list, unless ordered by size Produces the smallest leftover hole Worst-fit: Allocate the largest hole; must also search entire list Produces the largest leftover hole According to simulations, first-fit and best-fit are better than worst-fit in terms of speed and storage utilization. First-fit is generally faster.

Fragmentation External Fragmentation – when total memory space exists to satisfy a request, but the available spaces are not contiguous. Storage is fragmented into a large number of small holes caused by first-fit and best-fit strategies for memory allocation plus variable-partition scheme Internal Fragmentation – when allocated memory may be slightly larger than requested memory; this size difference is memory internal to a partition, but not being used First fit analysis reveals that given N blocks allocated, 0.5 N blocks are lost to fragmentation 1/3 may be unusable  known as the 50-percent rule

Fragmentation (Cont.) One solution to the external fragmentation : compaction Shuffle memory contents to place all free memory together in one large block Move all processes toward one end of memory All holes move in the other direction with one large hole of available memory produced. Expensive scheme Compaction is possible only if: relocation is dynamic and is done at execution time I/O problem Latch job in memory while it is involved in I/O Do I/O only into OS buffers Another solution to the external-fragmentation problem permitting the logical address space of the processes to be non-contiguous Segmentation and paging Now consider that backing store has the same fragmentation problems as main memory.

Segmentation Segmentation is a memory-management scheme that supports user (programmer) ’s view of memory Segmentation as a hardware supported memory mechanism: mapping the programmer’s view to the actual physical memory A program is a collection of segments A segment is a logical unit such as: main program procedure function method object local variables, global variables common block stack symbol table arrays

Programmer’s View of a Program Logical address space is a collection of segments. Each segment has a name and a length. <segment-number, offset> While Compiling, the compiler automatically constructs segments reflecting the input program. Since segments vary in length, memory allocation is a dynamic storage-allocation problem.

Logical View of Segmentation 1 4 2 3 1 2 3 4 user space physical memory space

Segmentation Hardware Logical address consists of a two tuple: <segment-number, offset> Segment number – an index to the segment table Offset – be between 0 and the segment limit, if it exceeds the segment limit, trap to OS Map two-dimensional programmer-defined addresses into one-dimensional physical addresses Mapping through a segment table – each table entry has: Segment base – contains the starting physical address where the segments reside in memory Segment limit – specifies the length of the segment The segment table is an array of base-limit register pairs. Segment-table base register (STBR) points to the segment table’s location in memory Segment-table length register (STLR) indicates number of segments used by a program; segment number s is legal, if s < STLR

Segmentation Hardware (Cont.)

Example of Segmentation Separate entry for each segment

Paging Physical address space of a process is allowed to be noncontiguous; process is allocated physical memory whenever available Avoids external fragmentation and the need for compaction. Basic method for implementing paging: Breaking physical memory into fixed-sized blocks called frames Dividing logical memory into blocks of same size called pages When executing a process, its pages are loaded into any available memory frames from their source (a file system or the backing store) To run a program of size N pages, need to find N free frames and load program Backing store is divided into fixed-sized blocks that are of the same size as the memory frames Keep track of all free frames Still have internal fragmentation

Address Translation Scheme Address generated by CPU is divided into: Page number (p) – used as an index into a page table which contains base address of each page in physical memory Page offset (d) – used as a displacement, combined with base address to define the physical memory address that is sent to the memory unit Page size (frame size) is defined by the hardware. Size is power of 2, between 512 bytes and 16 Mbytes per page. For given logical address space 2m and page size 2n High-order m-n bits of a logical address designates the page number n low-order bits designate the page offset Logical address structure: p is an index into the page table and d is the displacement within the page page number page offset p d m - n n

Hardware Support for Paging Every address generated by the CPU is divided into two parts: a page number (p) and a page offset (d) Page number is used as an index into a page table. Page table contains the base address of each page in physical memory.

Paging Model of Logical and Physical Memory

Paging Example 1 2 3 4 5 6 In the logical address, n=2 and m=4. Frame No. 1 2 3 4 5 6 7 In the logical address, n=2 and m=4. Using 32-byte physical memory and 4-byte pages

Paging (Cont.) Calculating internal fragmentation Page size = 2,048 bytes (n=11) Process size = 72,766 bytes 35 pages (71,680 bytes)+ 1,086 bytes Internal fragmentation of 2,048 - 1,086 = 962 bytes Worst case fragmentation = 1 frame – 1 byte On average fragmentation = 1 / 2 frame size So small frame sizes desirable. But overhead But each page table entry takes memory to track Disk I/O burden As a result, page sizes growing over time Solaris supports two page sizes – 8 KB and 4 MB

Paging (Cont.) Clear separation between the programmer’s view of memory and the actual physical memory. Programmer’s view: as one single space (in fact, scattered) The mapping is hidden from the programmer and is controlled by the OS. By implementation, the user process can only access its own memory. Page table includes only those pages that the process owns The OS must be aware of the allocation details of physical memory.

Free Frames Before allocation After allocation

Implementation of Page Table Page table for each process is kept in main memory. PCB contains a pointer to the page table like the other register values, the instruction counter. Page-table base register (PTBR) points to the page table. Page-table length register (PRLR) indicates size of the page table. In this scheme, every data/instruction access requires two memory accesses One for the page table and one for the data / instruction The two memory access problem can be solved by the use of a special fast-lookup hardware cache called associative memory or translation look-aside buffers (TLBs) Some TLBs store address-space identifiers (ASIDs) in each TLB entry – uniquely identifies each process to provide address-space protection for that process Otherwise need to flush at every context switch in order to prevent the next executing process from using the wrong translation information TLBs typically small (64 to 1,024 entries) On a TLB miss, value is loaded into the TLB for faster access next time Replacement policies must be considered Some entries can be wired down for permanent fast access (e.g. TLB entries for kernel code)

Associative Memory, TLB Associative memory – parallel search Address translation (p, d) If p is in associative register, get frame # out Otherwise get frame # from page table in memory Page # Frame #

Paging Hardware With TLB

Effective Access Time Associative Lookup =  time unit Hit ratio =  Can be < 10% of memory access time Hit ratio =  Hit ratio – percentage of times that a page number is found in the associative registers; ratio related to number of associative registers Consider  = 80%,  = 20ns for TLB search, 100ns for memory access Effective Memory Access Time (EAT) EAT = 0.80 x 120 + 0.20 x 220 = 140 ns Consider slower memory but better hit ratio ->  = 98%,  = 20ns for TLB search, 140ns for memory access EAT = 0.98 x 160 + 0.02 x 300 = 162.8 ns

Memory Protection Memory protection implemented by associating protection bit with each frame to indicate if read-only or read-write access is allowed Can also add more bits to indicate page execute-only, and so on Valid-invalid bit attached to each entry in the page table: “valid” indicates that the associated page is in the process’s logical address space, and is thus a legal page “invalid” indicates that the page is not in the process’s logical address space Or use PTLR (Page-Table Length Register) Any violations result in a trap to the kernel

Valid (v) or Invalid (i) Bit in a Page Table

Shared Pages Advantage of paging Shared code Private code and data Possibility of sharing common code Important in a time-sharing environment Shared code One copy of read-only (reentrant) code shared among processes (i.e., text editors, compilers, window systems) Similar to multiple threads sharing the same process space Also useful for interprocess communication if sharing of read-write pages is allowed Private code and data Each process keeps a separate copy of the code and data The pages for the private code and data can appear anywhere in the logical address space

Shared Pages Example

Structure of the Page Table Memory structures for paging can get huge using straight-forward methods Consider a 32-bit logical address space as on modern computers Page size of 4 KB (212) Page table would have 1 million entries (232 / 212) If each entry is 4 bytes  4 MB of physical address space / memory for page table That amount of memory used to cost a lot Don’t want to allocate page table contiguously in main memory Solution to this problem is to divide the page table into smaller pieces. Hierarchical Paging Hashed Page Tables Inverted Page Tables

Hierarchical Page Tables Break up the logical address space into multiple page tables A simple technique is a two-level page table We then page the page table

Two-Level Page-Table Scheme

Two-Level Paging Example A logical address (on 32-bit machine with 1K page size) is divided into: a page number consisting of 22 bits a page offset consisting of 10 bits Since the page table is paged, the page number is further divided into: a 12-bit page number a 10-bit page offset Thus, a logical address is as follows: where p1 is an index into the outer page table, and p2 is the displacement within the page of the inner page table Known as forward-mapped page table page number page offset p1 p2 d 12 10 10

Address-Translation Scheme

64-bit Logical Address Space Even two-level paging scheme not sufficient If page size is 4 KB (212) Then page table has 252 entries If two level scheme, inner page tables could be 210 entries * 4-byte Address would look like Outer page table has 242 entries * 4 bytes per entry = 244 bytes One solution is to add a 2nd outer page table But in the following example the 2nd outer page table is still 234 bytes in size And possibly 4 memory access to get to one physical memory location outer page inner page page offset p1 p2 d 42 10 12

Three-level Paging Scheme

Hashed Page Tables Common approach for handling address spaces larger than 32 bits with the hash value being the virtual page number The virtual page number is hashed into a page table Each entry in the hash table contains a chain (linked list) of elements hashing to the same location. Each element contains three fields: (1) the virtual page number (2) the value of the mapped page frame (3) a pointer to the next element in the chain (linked list) Virtual page numbers are compared in this chain searching for a match If a match is found, the corresponding physical frame is extracted

Hashed Page Table

Inverted Page Table Rather than each process having a page table and keeping track of all possible logical pages, track all physical pages One entry for each real page (or frame) of memory is added. Entry consists of the virtual address of the page stored in that real memory location, with information about the process that owns that page ASID (address-space identifier) in each entry of the inverted page table Only one page table is in the system. Simplified version of the inverted page table used in the IBM RT. Virtual address <process-id, page-number, offset> Inverted page-table entry <process-id, page-number> where process-id acts like an ASID If a match is found at entry i  physical address <i, offset> is generated. If no match is found  an illegal address access

Inverted Page Table (cont.) Decreases memory needed to store each page table with only one table, but increases time needed to search the table when a page reference occurs Because the inverted page table is sorted by physical address but lookups occur on virtual addresses  might need to search the whole table for a match Use hash table to limit the search to one — or at most a few — page-table entries TLB (Translation Look-aside Buffers) can accelerate access But how to implement shared memory? Shared memory is usually implemented as multiple virtual addresses (one for each process sharing the memory) that are mapped to one physical address. Because there is only one virtual page entry for every physical page.

Operation of an Inverted Page Table If a match is found at entry i  physical address <i, offset> is generated.

Segmentation Memory-management scheme that supports user view of memory A program is a collection of segments A segment is a logical unit such as: main program procedure function method object local variables, global variables common block stack symbol table arrays

User’s View of a Program

Logical View of Segmentation 1 4 2 3 1 2 3 4 user space physical memory space

Segmentation Architecture Logical address consists of a two tuple: <segment-number, offset> Segment number – an index to the segment table Offset – be between 0 and the segment limit, if it exceeds the segment limit, trap to OS Segment table – maps two-dimensional user-defined addresses into one-dimensional physical addresses; each table entry has: Segment base – contains the starting physical address where the segments reside in memory Segment -limit – specifies the length of the segment The segment table is an array of base-limit register pairs. Segment-table base register (STBR) points to the segment table’s location in memory Segment-table length register (STLR) indicates number of segments used by a program; segment number s is legal if s < STLR

Segmentation Hardware

Example of Segmentation Separate entry for each segment

Segmentation Architecture (Cont.) Protection With each entry in segment table associate: validation bit = 0  illegal segment read/write/execute privileges Protection bits associated with segments; code sharing occurs at segment level Since segments vary in length, memory allocation is a dynamic storage-allocation problem

Example: The Intel Pentium Supports both segmentation and segmentation with paging Each segment can be 4 GB Up to 16 K segments per process Divided into two partitions First partition of up to 8 K segments are private to process (kept in local descriptor table LDT) Second partition of up to 8K segments shared among all processes (kept in global descriptor table GDT) CPU generates logical addresses: <selector, offset> Given to segmentation unit which produces a linear address for each logical address The linear address is given to the paging unit, which generates the physical address in main memory The segmentation and paging units form the equivalent of the memory-management unit. Pages sizes can be 4 KB or 4 MB

Logical to Physical Address Translation in the Intel Pentium Logical-to-physical address translation in the Pentium Logical addresses: <selector, offset> s: segment number g: GDT/LDT p: protection 13 1 2 Selector Linear address (two level paging scheme)

Intel Pentium Segmentation

Pentium Paging Architecture <two level paging scheme for 4-KB page> (Linear Address) outer-most page table Pages sizes can be 4 KB or 4 MB. bypassing the inner page table

Linear Address in Linux Linux uses only 6 segments: kernel code kernel data user code user data task-state segment (TSS) : hardware context of each task during Context Switch, eg CR3 register default LDT segment Linux only uses two of four (2-bit) possible protection modes – kernel and user modes Uses a three-level paging strategy that works well for 32-bit and 64-bit systems Linear address in Linux broken into four parts: Linux rarely uses segmentation because it is designed to run on a variety of processors.

Three-level Paging Model in Linux What if using Linux on Pentium? Pentium only supports two-level paging. How Linux apply three-level paging on Pentium?  Quiz